【作者】 刘健；

【导师】 袁东风； Benny Bing；

【作者基本信息】 山东大学 ， 通信与信息系统， 2008， 博士

【摘要】 跨层设计思想是下一代无线通信系统的一项关键的理论创新,该方法打破了传统的分层设计思想,将原来被割裂的网络各层作为统一的整体进行设计、分析、优化和控制,同时充分利用物理层、MAC层以及其它各层之间强烈的相关性信息,进行无线网络协议性能的整体优化。本论文以跨层设计的思想为主线,研究了下一代无线通信网络的不同标准的系统跨层设计MAC协议的若干关键问题,并且通过理论和计算机仿真验证了跨层设计在吞吐量、丢包率和平均延时等参数相对于传统分层设计在性能上的显著优势。本论文的研究成果为下一代无线通信网络提供了可参考的跨层优化策略,而且还有助于为下一代无线通信系统的标准化提供参考。在本论文中,我们将跨层设计、非线性优化和资源动态分配作为主要的研究对象,通过建立包含各层参数的统一数学模型,使用非线性优化这个数学工具,将跨层设计和资源的动态分配有机的结合起来,从而达到提高网络性能的目的。具体来说,本论文主要研究工作和创新点如下:1、提出了一种新的基于跨层的链路自适应机制。首先详细研究了已有的链路自适应机制:自动率反馈机制(ARF)、基于接收端的自动率反馈机制(RBAR)、基于MAC层服务数据单元(MSDU)的自适应物理层模式选择机制和基于MAC层协议数据单元(MPDU)的链路自适应机制。通过比较,选择了目前性能最优的基于MAC层协议数据单元(MPDU)的链路自适应机制作为本工作的基础,它可以动态的调整MPDU的数据传输速率,从而使系统性能得到了大大的提升。基于上述分析,提出了一种基于IEEE 802.11a正交频分复用(OFDM)物理层和分布式协调(DCF)MAC层的跨层链路自适应机制。该机制通过检测网络负载信息和信道状态信息,动态的调整帧长、竞争窗口的大小和不同的物理层模式,从而显著的提高了网络的有效吞吐量(Goodput),并且减少了网络的丢包。2、利用非线性最优化的数学方法,先将理论分析得到的有效吞吐量的表达式作为目标函数,以帧长、竞争窗口的尺寸和物理层模式的选择作为限制条件,从而得到优化表达式。经数学证明,该非线性优化问题为凸优化,因此可以在凸优化理论的基础上展开,这样便可以使用凸优化的很多良好特性,方便该优化问题的求解。3、提出了一种基于IEEE 802.16e的跨层自适应子载波分配机制,称为跨层的无线资源分配机制(CL_ARRA)。这种机制工作于无线Mesh网络的基站(BS)中,即在下行链路中,使用IEEE 802.16e的正交频分复用多址(OFDMA)技术来动态的分配子载波,从而达到最大化系统吞吐量,满足服务质量的要求。该机制分为两步执行:首先基站通过信道状态信息(CSI)和队列状态信息(QSI)来动态的分配子载波,以获得子载波分配策略;然后通过已有的子载波分配策略,获得动态功率分配,并进一步获得下行链路的吞吐量。该机制不同于已有的其他方法,它可以工作在带有重叠区域的同频率的相邻小区中,使相邻的小区可以共享频率资源,同时它能使位于重叠区域中的移动节点发生冲突的概率降为最小,从而最大程度的提高系统的吞吐量并有效的降低系统的平均延时。4、基于上述无线Mesh网络的下行链路子载波分配机制(CL_ARRA),提出了一种新的冲突避免的上行MAC协议。该协议使用CL_ARRA分配的子载波资源,来对本小区的基站(BS)和移动节点(MS)进行访问。通过对上行链路传输机制的研究,提出了一种新的子载波退避算法,该算法动态的调整位于重叠区域中的移动节点的子载波的数量,从而达到冲突避免的目的。通过使用二维马尔可夫链(2-D Markov Chain)模型对该MAC协议进行分析,表明该协议能有效的降低位于重叠区域中的移动节点发生冲突的概率,从而提高系统的吞吐量。更多还原

【Abstract】 The idea of cross-layer design is a key theoretical innovation for next generation wireless communication systems.It discards the idea of layered design and views separate network layers as a whole to design,analyze,optimize,and control.At the same time,it makes full use of the closely related information between physical,MAC, and other layers to optimize the performance of wireless network protocol.With the idea of cross layer design,this thesis investigates several key problems in cross layer design of different protocols for next generation wireless communication networks. Through theoretical analysis and computer simulation,it verifies the advantage of cross-layer design over conventional layered design in throughput,packet loss,and average delay.The research results provide next generation wireless communication networks reference cross-layer optimization strategies and are helpful for the standardization.In this thesis,we mainly focus on cross-layer design,non-linear optimization,and dynamic resource allocation.We build the general mathematical model with parameters from different layers,and use non-linear optimization as a mathematical tool to combine cross-layer design and dynamic resource allocation together so as to improve the network performance.Our work and contributions include:1.We propose a new link adaptation mechanism based on cross-layer design.First we investigate the current link adaptation mechanisms including ARF,RBAR,MSDU based adaptive physical layer mode selection,and MPDU based link adaptation.We choose MPDU based link adaptation as the basis of our work,which is with best performance through comparison.It dynamically adjusts the data transmission rate of MPDU so that the system performance improves greatly.Based on the above analysis,we propose a cross-layer link adaptation based on IEEE 802.11a OFDM physical layer and DCF MAC layer.By detecting network load information and channel state information,it dynamically adjusts the frame length,contention window length,and physical layer mode so as to improve the goodput and reduce the network packet loss. 2.With the mathematical tool of non-linear optimization,we use the theoretical expression of goodput as the objective,and frame length,contention window length, and physical layer mode as the constraints.It is a convex optimization problem according to the proof.Therefore,based on the convex optimization theory with good characteristics,the solution is easily found.3.We propose a cross-layer adaptation subcarrier allocation,which is called CL_ARRA.Working in the base station(BS) of wireless mesh network,i.e., downlink,it uses IEEE 802.16e OFDMA technique to dynamically allocate the subcarriers to maximize system throughput and satisfy the quality of service requirement.It includes two steps:first the base station dynamically allocates the subcarriers according to CSI and QSI so as to obtain the strategy of subearrier allocation;then with the subcarrier allocation,it acheives dynamic power allocation to improve the throughput of downlink.Different from other methods,this scheme works in overlapped neighborhood cells using the same frequency.It allows neighborhood cells to share frequency resources and minimize the collision probability of mobile nodes in the overlapped region.Therefore,it improves the system throughput and reduces the average delay effectively.4.Based on the above downlink subcarrier allocation mechanism(CL_ARRA),we propose a new collision avoidance uplink MAC protocol,which uses the subcarrier resource allocated by CL_ARRA to access the BS and MS in the cell.Through the investigation on uplink transmission mechanism,we introduce a new subcarrier backoff algorithm,which dynamically adjusts the quantity of subcarriers of mobile nodes in the overlapped region to avoid collision.Using 2-D Markov Chain model, the analysis of the MAC protocol shows that the protocol effectively reduces the collision probability of mobile nodes in the overlapped region and improves the system throughput.更多还原